Paul Industries designs and installs ultrapure water systems and high-purity utilities for Oregon’s semiconductor and advanced manufacturing sector. The single most important thing to understand about ultrapure water is counterintuitive: it is aggressive because it is pure. Strip everything out of water and what remains is a powerful solvent that will dissolve whatever it is stored in and piped through, and the system’s entire design follows from managing that.
Request a quote or call 201-450-8280
Purity is a state that has to be maintained continuously
Oregon’s Silicon Forest is a genuine industrial cluster rather than a marketing phrase. Hillsboro hosts Intel’s largest and most comprehensive site in the world for research and manufacturing, and semiconductors are Oregon’s largest manufacturing sector, employing roughly 33,000 people across the state.
Every one of those fabs runs on ultrapure water, and the volumes are large. The specification is not a single number, though resistivity is the one people quote. A resistivity of 18.2 megohm-centimeter at 25 degrees Celsius is the theoretical maximum for pure water and it indicates the absence of ionic species. It says nothing about organics, nothing about particles, nothing about dissolved gases and nothing about trace metals, all of which are specified separately and all of which matter to a process working at the scales modern semiconductor manufacturing works at.
So the system is not a treatment plant with a pipe on the end. It is a continuously operating loop whose job is to hold a condition that the water itself is constantly trying to lose.
Three things attack that condition and the design answers each of them.
The water dissolves its container. Ultrapure water has no dissolved load, so it takes material from whatever it contacts. Stainless steel, excellent for pharmaceutical water, leaches metals into ultrapure water at levels that matter here. That is why high-purity distribution is built in fluoropolymers, typically PVDF or PFA, and why the joining method is thermal fusion rather than welding of metal.
Biology establishes anywhere it can. Ultrapure water is a poor growth medium and it is not sterile. Biofilm establishes wherever water is slow or stationary, and once established it sheds organics and particles continuously into the loop. Continuous circulation at velocity with no dead legs is the primary control, and it is a design property rather than an operating procedure.
Atmosphere gets in. Dissolved oxygen and carbon dioxide arrive through any point where the loop meets air, and carbon dioxide in particular degrades resistivity directly. Tanks blanketed, vents filtered, and the loop kept closed.
The loop, the polish and the point of use
| Stage | Target | Failure signature |
|---|---|---|
| Pretreatment | Bulk solids, hardness, chlorine | Downstream membranes fouling early |
| Reverse osmosis | The large majority of dissolved load | Rising conductivity into the next stage |
| Degasification | Carbon dioxide and oxygen | Resistivity that will not reach target |
| Ion exchange and EDI | Remaining ionic species | Resistivity falling at the polish outlet |
| Ultraviolet | Organics and microbial control | Rising TOC |
| Final filtration | Particles at the point of use | Particle counts at the tool |
The architectural point worth making is that the polish loop matters more than the makeup train. A makeup system can be rebuilt or supplemented; the distribution loop is embedded in the building and is what actually determines what arrives at a tool. Money spent on loop design, velocity, dead leg elimination and point-of-use polishing buys more than money spent making the front end slightly better.
The corresponding design rule is that every branch to a tool is a potential dead leg, and the plant will add tools. A loop designed with generous provision for future take-off points, arranged so that a disused branch can be removed rather than capped, avoids a slow accumulation of stagnant stubs that degrade the whole system over a decade.
| Continuous load | Per year | Over ten years |
|---|---|---|
| 50 kW | $35,259 | $352,590 |
| 100 kW | $70,518 | $705,180 |
| 200 kW | $141,036 | $1,410,360 |
Oregon is worth a note here because its position surprises people. Despite substantial hydroelectric generation, Oregon industrial electricity averages 8.05 cents per kilowatt-hour, essentially at the 8.13 cent national average (EIA, 2024), and materially above neighbouring Washington. Anyone assuming Pacific Northwest power is uniformly cheap and sizing an energy-intensive system on that assumption should check the actual tariff first.
Reclaim, which is where the operating cost sits
A fab uses a great deal of water and it does not use all of it equally hard. Rinse water that has touched almost nothing leaves the process barely changed, and reclaiming it is standard practice rather than an innovation.
The engineering question is segregation, and it has to be answered at design stage because it is about plumbing. Streams have to be separated at source by the contamination they carry, so that the clean rinse stream can be reclaimed economically rather than being mixed with a stream containing something difficult and thereby made expensive to treat. A fab that combines its drains has decided not to reclaim, whether or not anyone framed it that way.
The second question is where the reclaimed water goes. Returning it to the front of the ultrapure train is one option and it is the most demanding. Using it for cooling tower makeup, scrubber water, or other utility duties is far easier to justify and frequently captures most of the available benefit for a fraction of the treatment cost.
We design and install the ultrapure distribution loops, point-of-use polishing, reclaim segregation and treatment, and the interfaces to tools, and we coordinate with makeup system suppliers rather than pretending to be one. Fluoropolymer systems are fusion joined with documented parameters, and where stainless is appropriate elsewhere in the plant we build to ASME BPE with orbital welding to AWS D18.1 and passivation to ASTM A967 after cleaning per ASTM A380.
Standards referenced: EIA electricity price data · ASME BPE · ASTM A967 · ASTM A380
Frequently asked questions
Do you build ultrapure water systems in Oregon?
Yes, across Hillsboro, Portland, Beaverton, Corvallis, Gresham and statewide: ultrapure distribution loops, point-of-use polishing, reclaim segregation and treatment, tool interfaces, and the supporting high-purity utilities around them.
Why is ultrapure water corrosive?
Because it has no dissolved load, which makes it an aggressive solvent. It takes material from whatever it contacts, so it dissolves its own container and piping. That is the central design constraint and it is why materials that suit pharmaceutical water do not necessarily suit this duty.
Why not use stainless steel?
Because stainless leaches metals into ultrapure water at levels that matter at semiconductor process scales. High-purity distribution is normally built in fluoropolymers such as PVDF or PFA, joined by thermal fusion rather than by welding metal, with the fusion parameters documented.
Is resistivity the whole specification?
No, though it is the number most often quoted. A resistivity of 18.2 megohm-centimeter at 25 degrees Celsius is the theoretical maximum and indicates the absence of ionic species. It says nothing about organics, particles, dissolved gases or trace metals, all specified separately and all consequential.
What controls biological growth?
Continuous circulation at velocity with no dead legs, supported by ultraviolet treatment. Ultrapure water is a poor growth medium but it is not sterile, and biofilm establishes wherever water is slow or stationary, then sheds organics and particles into the loop continuously.
Which matters more, the makeup train or the loop?
The loop. A makeup system can be rebuilt or supplemented, while the distribution loop is embedded in the building and determines what actually arrives at a tool. Money spent on loop design, velocity, dead leg elimination and point-of-use polishing generally buys more than improving the front end.
How should we plan for future tools?
With generous provision for future take-off points, arranged so a disused branch can be physically removed rather than capped. Every branch is a potential dead leg, plants always add tools, and capped stubs accumulate quietly into a system-wide degradation over a decade.
Is water reclaim worth doing?
Yes, and it is standard practice. The engineering question is segregation, which has to be decided at design stage because it is about plumbing. A fab that combines its drains has effectively decided not to reclaim, whether or not anyone framed the decision that way.
Is Oregon power cheap?
Less than people assume. Despite substantial hydroelectric generation, Oregon industrial electricity averages 8.05 cents per kilowatt-hour, essentially at the 8.13 cent national average and materially above neighbouring Washington (EIA, 2024). Anyone sizing an energy-intensive system on assumed cheap Northwest power should check the actual tariff.
How do I get a quote for an Oregon ultrapure water project?
Use the form on this page or call 201-450-8280. Useful inputs are flow requirements and the specification beyond resistivity, the number and type of points of use, whether reclaim is in scope, feed water quality, and whether this is a new build or an extension of an existing loop.
How is an ultrapure water specification written for a specific process node?
From the process's sensitivity to each contaminant class, taken from the tool suppliers' requirements and industry guidance for the node, with limits set for resistivity, total organic carbon, particles, dissolved oxygen, silica, boron and metals individually. A single resistivity number is not a specification for a fab.
What does the polishing loop do?
It continuously recirculates water through mixed-bed ion exchange, ultraviolet oxidation for organics, degasification and final ultrafiltration, so that water reaching the tools is repolished on every pass. The loop is where the final quality is made, and it runs continuously.
How is total organic carbon reduced to parts-per-billion levels?
By ultraviolet oxidation at 185 nanometres, which breaks organics into carbon dioxide and ions removed by downstream ion exchange, combined with reverse osmosis upstream. TOC is measured online because it is the parameter most sensitive to upsets.
How are dissolved oxygen and other gases controlled?
By membrane degasification and vacuum degassing in the loop, because dissolved oxygen affects wafer surfaces and dissolved gases affect measurements. Gas control is a separate stage from ion removal.
Which polymers are used in an ultrapure water loop and why?
High-purity PVDF for the distribution loop and PFA for the point-of-connection tubing, chosen because they do not leach ions or organics, do not corrode in ultrapure water and can be fusion-welded without crevices. Valve and fitting materials follow the same rule, and every material is qualified for extractables.
What keeps bacteria out of an ultrapure loop?
Continuous flow with no dead legs, ultraviolet sterilisation and final filtration, low-nutrient water with total organic carbon in the parts-per-billion range, and periodic sanitisation with hydrogen peroxide or ozone where the materials permit. The loop is designed so that bacteria have nowhere to attach and nothing to eat.
How is silica controlled in ultrapure water?
By reverse osmosis and ion exchange, with reactive silica measured at parts-per-billion levels because it deposits on wafers. Oregon's volcanic groundwater can carry silica that has to be managed from the make-up train onward.
How is a fab's reclaim water integrated with its ultrapure demand?
By segregating the rinse water streams that are clean enough to reclaim, treating them to a quality the make-up train accepts, and blending them into the feed at a rate the train's design allows, so that reclaim reduces raw water draw without loading the polishing loop. Reclaim is a make-up strategy, not a loop strategy.
How long does UPW commissioning take?
Weeks to months, because the loop has to be flushed, sanitised and polished until the quality parameters settle, and every material in the system releases contaminants that have to be washed out. It is the longest single commissioning activity in a fab and is planned as such.
How is a new tool tied into a running ultrapure loop?
From a pre-installed valved stub, with the branch fabricated, flushed and qualified before the valve is opened, and with the loop's flow and pressure rebalanced to include the new draw. Cutting into a live loop is avoided because the loop then has to be requalified in full.
What monitoring does a UPW system run?
Resistivity, TOC, particles, bacteria, dissolved oxygen, silica and specific ions, online where possible and by sampling otherwise, with the data trended. The monitoring is a large part of the system's operating cost.
What make-up water sources do Oregon fabs use?
Municipal supply, which in the Portland area is soft and low in dissolved solids, which simplifies the front end. Surface-derived supplies vary seasonally in organics and turbidity, which the pretreatment handles.
How is UPW heated or cooled for tools?
Some tools need hot ultrapure water, produced by heat exchangers on the loop with materials that do not contaminate. Temperature control adds complexity and is provided only where the tool needs it.
Do smaller Oregon manufacturers need ultrapure water?
Photonics, electronics and some materials plants need high-purity water short of full UPW, and the same design principles apply at smaller scale with less monitoring. The specification follows the process.
What is the commonest UPW problem?
A material or component in the loop that was not ultrapure-compatible, leaching organics or ions and holding the loop off specification through months of commissioning. Material control at construction is the prevention.
Commissioning: the part of the schedule nobody believes until they live it
A newly built ultrapure water loop does not produce ultrapure water. It produces water that is gradually becoming ultrapure, and the gap between those two states is measured in days to weeks of continuous recirculation.
The reason is the same one that governs the whole system. Every internal surface in a new loop carries something: extrusion residues from the pipe, material from the fusion joints, particles generated during installation, handling contamination from fabrication. Ultrapure water dissolves and entrains all of it, which is exactly what you want it to do, and the loop cannot reach specification until the surfaces have given up what they have to give.
The practical consequences are worth stating plainly, because they are routinely underestimated at the planning stage.
Flushing is a phase, not a step. The loop runs at full velocity, continuously, to drain or through the treatment train, while the parameters are tracked. Resistivity typically arrives first and is the least informative. TOC and particle counts take longer and are the ones that actually govern release. A loop that has hit resistivity target but is still shedding particles is not ready, and reading a single parameter is how a loop gets released too early.
It consumes a lot of water. Extended flushing of a large loop uses a serious volume, and where the site’s supply or discharge is constrained that becomes a logistics problem. It is worth establishing early whether the flush water can be recovered into a lower-grade duty rather than sent to drain.
Installation cleanliness directly shortens it. This is the lever worth pulling. Piping fabricated and stored clean, joints made in controlled conditions, open ends capped whenever work stops, and no site contamination allowed into the system all reduce the flush duration substantially. A fortnight of installation discipline can save a fortnight of commissioning, and commissioning time is on the critical path in a way installation time often is not.
Sample where it matters. The polish outlet tells you what the treatment train is doing. The points of use tell you what the tools will actually receive, and those are different numbers, particularly on a long loop or one with many branches. Release criteria should be written against point-of-use data.
We plan the flush as a scheduled phase with defined parameters and release criteria rather than as an open-ended activity, because an undefined flush has a habit of being ended by schedule pressure rather than by data.
Planning or extending an Oregon ultrapure water loop?
Tell us your specification beyond resistivity and how many points of use you expect to add. The second answer shapes the loop more than the first. Call 201-450-8280 or use the form below.